International Journal of Impact Engineering 36 (2009) 1250–1258

Contents lists available at ScienceDirect

International Journal of Impact Engineering
journal homepage: www.elsevier.com/locate/ijimpeng

Energy absorption during projectile perforation of thin steel plates and the kinetic energy of ejected fragments
J. Dean a, C.S. Dunleavy a, P.M. Brown b, T.W. Clyne a, *
a b

Department of Materials Science & Metallurgy, Cambridge University, Pembroke Street, Cambridge CB2 3QZ, UK DSTL, Physical Sciences Research Department, Salisbury, UK

a r t i c l e i n f o
Article history: Received 21 August 2008 Received in revised form 2 May 2009 Accepted 10 May 2009 Available online 21 May 2009 Keywords: Ballistic Impact Perforation ABAQUS Johnson & Cook

a b s t r a c t
This paper concerns energy absorption in thin (0.4 mm) steel plates during perforation by spherical projectiles of hardened steel, at impact velocities between 200 and 600 m sÀ1. Absorbed energies have been obtained from measured incident and emergent projectile velocities. These tests were simulated using ABAQUS/Explicit, using the Johnson and Cook plasticity model. A strain rate-dependent, critical plastic strain fracture criterion was employed to model fracture. Good agreement is obtained between simulations and experiment and the model successfully captures the transitions in failure mode as projectile velocity increases. At velocities close to the ballistic limit, the plates fail by dishing and discing. As the incident velocity is increased, there are two transitions in failure mode, firstly to shear plugging and secondly to fragmentation and petalling. The simulations also show that, during the latter mode of failure, the kinetic energy of ejected debris is significant, and failure to include this contribution in the energy balance leads to a substantial over-estimate of the energy absorbed within the sheet. Information is also presented relating to the strain rates at which plastic deformation occurs within the sample under different conditions. These range up to about 105 sÀ1, with the corresponding strain rate hardening effect being quite substantial (factor of 2–3 increase in stress). Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction Perforation of plates during projectile impact is a complex process, commonly involving elastic and plastic deformation, strain and strain rate hardening effects, thermal softening, crack formation, adiabatic shearing, plugging, petalling and even shattering. These effects depend on the properties and geometries of projectile and target and on the incident velocity. There have been many theoretical and experimental studies in this area, which are covered in the comprehensive reviews of Corran et al. [1], Anderson and Bodner [2] and Corbett et al. [3]. Specific studies have been undertaken on petalling phenomena [4,5], shear plugging failure [6,7] and dishing [8] in thin metallic plates, while the effects of target thickness, impact obliquity [9] and projectile nose shape [10,11] have also been studied. The main failure modes are ductile hole enlargement (lateral displacement of material), petal formation, plugging failure via through-thickness shear (often accompanied or preceded by adiabatic heating), dishing, which is characterised by large plate

* Corresponding author. E-mail address: twc10@cam.ac.uk (T.W. Clyne). 0734-743X/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijimpeng.2009.05.002

deflection, membrane stretching and tensile tearing (discing) and, finally, fragmentation. It is not uncommon for a combination of these failure modes to operate simultaneously. Furthermore, the failure mode often changes with increasing projectile velocity [12]. The Finite Element Method (FEM) has become an increasingly useful tool for the analysis of impact events, and current commercial codes are well suited to dealing with problems involving large deformations and elevated strain rates. The necessary computational resources are also routinely available. However, as highlighted by Zukas [13], erroneous results can be obtained. Common errors include the use of models that are ill-suited to the problem, failure to recognise numerical instabilities (which may be attributed to physical phenomena), poor mesh specification, and use of inappropriate property data and constitutive relations. Of these, the latter are amongst the most common and, potentially, most significant. The constitutive behaviour of most metals is, in general, fairly well understood and standard expressions are routinely implemented into FEM codes. These normally incorporate the effects of strain, strain rate and temperature on the effective stress. Examples include the phenomenological model of Johnson & Cook [14] and the more theoretical, dislocation dynamics-based models of Zerilli and Armstrong [15]. The Johnson & Cook plasticity model is well

particularly at high velocities. impacted specimens were sectioned using EDM and photographed. from the stainless steel sheets. If the associated thermal softening exceeds the strain hardening. / International Journal of Impact Engineering 36 (2009) 1250–1258 1251 suited to high rate deformation. to simulate penetration experiments on 1 mm thick Al sheets. a high speed framing camera was employed (since a shorter exposure time was required). Circular disk specimens were cut. The sensitivity of the predicted results to mesh density was a key concern during the simulations (particularly since large strain gradients were expected).17. for example. the incident velocity was measured using a series of light gates. such that a circle of 60 mm diameter was exposed.17] on projectile penetration of steel and aluminium sheets. Their simulations agree reasonably well with experimental data. then continued loading is likely to result in failure at stress levels well below the static strength of the material at ambient temperature [20. is that of Yungwirth et al. As impact velocities were increased beyond the ballistic limit. Over the critical period (w170 ms). critical plastic strain criterion is employed. the fracture constants (D1. [24]. Numerical modelling 3. Friction between the projectile and the plate was taken as negligible. and mass losses (due to the ejection of plugs or multiple fragments) were calculated. that a relatively fine mesh size is often required in order to capture their formation. using Johnson & Cook constitutive relations. linear quadrilateral shell elements – type S4R. These are bands of localised plastic shear strain. over the impact velocity range covered here. which has been widely employed [10.5 mm diameter. were rigidly clamped around the periphery. isotropic strain hardening. which is insensitive to further refinements beyond 2457 elements. [10] also used a Johnson & Cook elastic–viscoplastic model. of 76. they are typically 10–100 mm in width.22. At low projectile velocities (170–250 m sÀ1). Teng and Wierzbicki [18] found that those based on accumulated equivalent plastic strain were most reliable. since they are usually more reliable [13]. At intermediate projectile velocities (w250–350 m sÀ1). In the current study. separated by a known distance. Adiabatic shear band formation has been modelled numerically [7. so local mesh dimensions need to be correspondingly fine. based on measurement of the delay between the generation of current in two copper coils. who showed experimentally that sliding friction typically absorbs less than 3% of the total incident energy.2. The experimental work in the present study concerns 304 stainless steel. yielding. strain rate hardening and softening due to adiabatic heating. The projectile velocity was calculated from the inter-frame time and the distance travelled. capturing 12 sequential images on Polaroid film. and in general is reported to capture the main features of penetration and perforation – an example being the studies of Borvik et al. 12 polaroid images were captured.19] for simulating the perforation of metallic plates. Penalty contact was defined between the projectile and the plate. After penetration. . as well as ballistic limit velocities. Strain was measured using a linear variable displacement transducer (LVDT). Table 3 shows the predicted absorbed energy as a function of mesh density and minimum element size for an impact velocity of 600 m sÀ1. Specimens. and there exists a threshold mesh density value for solution convergence.1). To the authors’ best knowledge. who examined the ballistic performance of sandwich panels with pyramidal lattice cores and compared it with that of monolithic plate of equal areal density.1. comprising three light emitting diodes and three light receiving photodiodes. using EDM. Stress–strain responses were recorded and material property data were inferred. The projectile was modelled as an analytical rigid body and simply assigned mass. In steels. they observed a small decrease in the specific absorbed energy by the monolithic plates. The exposure time for all of the images was 1 ms and the inter-frame times ranged from 10 to 70 ms. Gupta et al. ´ An encastre boundary condition was specified to simulate the clamping conditions. Specimens were impacted at normal incidence. Their model included the effects of linear thermo-elasticity. that shear bands are commonly expected if thermal softening is taken into account. from 0. These can be either instantaneous. A further potential complication at high strain rates is the possible formation of adiabatic shear bands. The high speed photographic images were taken with an Ultranac FS501 image converter camera. despite their complexity. Dean et al. This assumes that projectile deformation is negligible and observations of projectiles after impact confirmed that there was no noticeable plastic deformation.4 mm thick sheets of grade-304 stainless steel that had been annealed in inert atmosphere at a temperature of 1195  C for 1.. incident and residual velocities were measured using high speed video equipment (Photo-sonics Phantom V4. [7. At high projectile velocities (w350–600 m sÀ1). Mesh formulation and boundary conditions The steel sheets were modelled as elastic–plastic shells and meshed with reduced integration. followed by a monotonic increase.21]. The inter-frame and exposure times were in the ranges 10–36 ms and 6–9 ms.D5) are more difficult to obtain. with a range of incident velocities.01 sÀ1.5 h. Tensile tests were conducted using a 10 kN ESH servo-hydraulic testing machine at a strain rate of 0.) However.3 high speed video camera).16. It can be expressed in the form !# " h ðd3=dtÞpl À Ái * 3f . For example. The shell element formulation is well suited to dynamic analyses involving large plastic bending strains [25]. with high mesh densities and when using an adaptive remeshing algorithm. which requires relatively few parameters to be evaluated.23]. by spherical projectiles of hardened steel. In a comparative study of six fracture conditions. when the moving (magnetic) projectile passes through them. Chou et al. They noted.pl ¼ D1 þD2 exp D3 s ð1þD5 qÞ 1þD4 ln ðd3=dtÞ0 (1) (A nomenclature listing is included in this paper as an Appendix. 8 mm in diameter and of 2 g mass. the only previous study on this material. Specimens were weighed before and after testing. using the Photo-sonics Phantom software. Experimental procedures Tensile test specimens were cut using Electric Discharge Machining (EDM). One such fracture criterion is that of Johnson & Cook. 2. The residual velocity was measured using an electromagnetic induction technique. This fracture criterion was coupled with the Johnson and Cook plasticity model (Section 3.16. The projectiles were back-illuminated.. Simulations of this type commonly employ fracture criteria. time-dependent or micromechanical [13].J. whilst the Johnson & Cook plasticity parameters are relatively easy to determine from simple mechanical tests (at least for low strain rates). which is consistent with the results of Krafft [26]. They successfully predicted the residual velocities of penetrating projectiles. Multiple tests must be conducted over a wide range of strain rate and stress triaxiality. Micromechanical models are often preferred. within which large temperature rises can occur. plastic flow. using a Bowens flash (with a light duration of several ms). however. including a simple critical fracture strain. [23] showed. The mesh density was refined in the region directly beneath the impact site. 3. a strain rate-dependent.

however. B and n are strain hardening parameters. which defines the flow stress as a function of equivalent plastic strain. it seems likely that they are at least broadly appropriate for the present work.07 A (MPa) 310 (d3/dt)0 (–) 0.95 . Tmelt is the melting temperature and Ttransition is the transition temperature defined as the one at or below which there is no temperature dependence on the expression of the yield stress. since these data [27] are reportedly [28] representative of annealed 304 stainless steel. i. in the parametric study conducted by Borvik et al. which were obtained from Lee et al. [29]. are listed in Table 1.pl is the plastic strain to failure. Fig. The base level (maximum) triaxiality can be calculated using Bridgman’s analysis [30]. The values of the Johnson & Cook parameters used in the present work. n. and thus characterises the degree to which plasticity is being promoted.3. of course. Extraction of global information about strain rate distributions For each volume element. The stress triaxiality is the ratio of the mean stress to the equivalent stress. and following the work of Lichtenfeld et al.2. This algorithm is relatively crude. [27]. the average strain rate at which deformation occurred was recorded. during each time interval. using parametric values obtained by Radford et al. Fracture A strain rate-dependent. and no such tests were conducted during the current work. However. as was the strain induced and the plastic work done. 3D. Instead. once initiated. can be treated as a shell. It is also worth noting that. but it was ensured that the time interval of the computations was short r (kg mÀ3) 7800 Tm (K) 1673 E (GPa) 200 Tt (K) 293 3f (–) 0. Material response characterisation 3. based on the Johnson–Cook plasticity relation. [33] on type-304 stainless steel. This is a plausible approximation. The damage. the fracture strain was assumed to decrease by 20% (compared to the quasi-static value – measured experimentally) for a strain rate of 104 sÀ1. [34] and fracture strain information given by Lichtenfeld [33]. which is not strictly true. D. accumulates in an element according to the following relation: D ¼ X D3pl 3f $pl (4) where D3pl is the increment of accumulated equivalent plastic strain and 3f. the stress triaxiality was nearly insensitive to strain rate and inertia. Johnson & Cook plasticity The Johnson & Cook plasticity formulation.pl ððd3=dtÞpl .1252 J. although to determine this parameter over a range of strain rate and temperature requires a significant number of tests [31]. [27]. is assumed to evolve linearly until the material is fully degraded and can no longer sustain load. This isn’t entirely satisfactory. The equivalent plastic strain at the onset of fracture 3D$pl is assumed to be a function of the equivalent plastic strain rate and stress triaxiality.2.65 cp (J kgÀ1 KÀ1) 440 m (–) 1 a (–) 0. m and C are material parameters and q is the non-dimensional temperature given by: 0 b q hðT À Ttransition Þ=ðTmelt À Ttransition Þ 1 T < Ttransition Ttransition T > Tmelt (3) T Tmelt where T is the current temperature. For cases in which the strain rate varied during the time interval beyond the range covered by a single bin. with plane stress conditions satisfied. Dean et al. once this condition is satisfied in an element. was employed in all simulations The dynamic flow stress is expressed by the following relation [14] sd ¼ A þ B 3pl h À Án i " 1 þ C ln ðd3=dtÞpl ðd3=dtÞ0 !# À 1Àq mÁ (2) where sd is the dynamic flow stress. SÞ. critical plastic strain-based fracture criterion was employed. A stress triaxiality of 1/3 does.2. The constant A is the yield stress under quasi-static conditions. m controls the temperature dependence and C the strain rate dependence. The damage.1. 3pl is the equivalent plastic strain. from the initial notch diameter. assuming that the plate material Table 1 Johnson & Cook plasticity and thermal property parameters for 304 stainless steel [27]. B.e. Stress–strain relationships for 304 steel.2. imply that the stress state of the plate is dominated by uniaxial tension. assume that the material used in this study is similar to that of Lee et al. 3. strain rate and temperature. the energy was partitioned between the bins concerned using a proportional algorithm. it is removed from the mesh. it has previously been reported that [29] Bridgman’s theory can lead to inaccurate estimates of the stress triaxiality. These increments of energy were accumulated into a series of bins covering specific strain rate ranges. a triaxiality of 1/3 was assumed. 1. However. 1. The stress triaxiality was also relatively insensitive to temperature. since high strain rate fracture data were unavailable. In any case.2. 1. A progressive reduction in fracture strain was assumed over the strain rate range. (d3/dt)0is a reference strain rate. and the data are plotted in Fig. Failure occurs when D ! 1. which is known to affect the likelihood of fracture – as the triaxiality increases. / International Journal of Impact Engineering 36 (2009) 1250–1258 3. as shown in Fig. The fracture strain is established as a tabular function of strain rate. since microstructural differences could affect the mechanical properties. A.01 B (MPa) 1000 n (–) 0. the strain to fracture decreases [29]. This does. based on the work of Zhou [32]. ðd3=dtÞpl is the equivalent plastic strain rate.33 C (–) 0. 3.

114 0 0 0 0 0 0 0 0 1.5 16.9 187. ðd3=dtÞ.9 5.7 18.5 190.3 19.3 20. It was obtained by simply multiplying each strain rate bin upper boundary by the corresponding fraction of energy absorbed within that strain rate bin.91 35.863 41. petalling.2 184. For an individual element.913 41.73 217. i.3 127.53 544. associated fragment kinetic energy (obtained assuming that all fragments acquire the residual projectile velocity) and the energy absorbed within the specimen (difference between energy loss of projectile and kinetic energy of fragments).7 18. (6)).0 17. The energy absorbed in the plate is expected to be (predominantly) dissipated via plastic deformation and fracture. 2(b) shows the experimental data of Fig.) A ‘‘characteristic strain rate’’.5 580 583 592 109.50 32.38 10. A comparison is shown in Fig.1 40.5 188.58 29.54 8. dishing.2 181.1 6.3 19. loss of specimen mass after impact.3 179.3 214.1 19.9 14.237 0.09 290.6 15. Dean et al. The main focus of the present study is on the energy absorbed by the plate. for the residual velocity as a function of the incident velocity.9 121.J.1. re-plotted as absorbed energies (difference between incident and emergent projectile kinetic energies).6 286. excluding the kinetic energy associated with expelled fragments (Eq. 4.5 172. such that fragment expulsion starts to become significant.250 0. and the associated deformation and fracture mechanisms. 2(a) between experimental data and predictions.3 15.4 14.43 545.9 14.71 0 6. (As with other aspects of the computation. 2. / International Journal of Impact Engineering 36 (2009) 1250–1258 1253 enough to ensure that the associated error was small.8 36. The table includes incident and residual projectile velocities. It can be seen that. No. 2(a).0 17.4 172.0 41. It can be seen that agreement with experiment is good over the complete range of (sub-sonic) incident velocities being investigated here.3 0 0 0 0 0 0 0 0 0.8 14.822 Fig.36 45.075 41.4 Hinged cap 4 Petals Hinged cap Hinged cap Hinged cap 6 Petals 7 Petals 4 Petals Plugging Hinged cap Plugging/petalling Plugging/petalling Plugging/petalling Petalling/fragments Petalling/fragments Petalling/fragments Petalling/fragments Fig.5 10.2 17. .98 560.156 0.47 19. plugging or fragmentation.e. of elements 220 585 1452 2457 4516 5818 6602 Minimum element size (mm) 815 433 132 120 83 41 32 Absorbed energy (J) 49.34 17. Also plotted in Fig. corresponding loss of projectile kinetic energy (DU). Comparison between measured and predicted dependence on incident velocity of (a) residual velocity and (b) absorbed energy.3 16.24 541.4 7.199 0. representing the mean strain rate throughout the specimen for a given incident velocity. discing. was calculated by considering all volume elements.21 292. Model predictions and comparisons with experimental data 4. Run Vin (m sÀ1) Vres (m sÀ1) DU (J) mfrag (g) Ufrag (J) Uplate (J) Failure mode 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 176.098 0 0. it was confirmed that the meshing and time interval selection were sufficiently refined to ensure that they had no significant effect on the predictions. the energy dissipated is given by: 3 Z 3f 0 Table 3 Mesh sensitivity analysis showing the predicted absorbed energies for an impact speed of 600 m sÀ1.e.2 6.5 178.43 42.3 20. based on Johnson–Cook input data (with and without the kinetic energy of specimen fragments being included in the predicted plots).8 12.94 19.202 0.4 18. It can be seen that the level of agreement is very good. bulging. TÞd3 (5) so that the total work done is equal to Uplate ¼ n X 1 Uelement (6) Table 2 Measured projectile velocities and mass losses.9 220.7 39.1 7. A summation was then made for all bins. across the complete time domain. weighted by the fraction of energy absorbed in a given element during a given time step.2 17. 2(b) is the predicted energy absorbed within the specimen by plastic deformation and fracture – i. inferred energy changes and observed failure modes for individual runs.1 19.7 222.222 0. obtained using the Johnson and Cook plasticity formulation. as expected.3 228 310 316 318 576.9 121. At an incident velocity of the Uelement ¼ sð3.89 43. this contribution to the absorbed energy becomes significant at higher incident velocities (>w300 m sÀ1).8 112. Effects of projectile incident velocity A summary is shown in Table 2 of the experimental data obtained from each run.

Fig. Model predictions. of the progression of fracture and the surface stress field. and (iii) fragmentation. viewed from the rear of the specimen. Fig. Fig. The failure modes observed during these experiments are: (i) dishing and discing. and the transitions in failure mode with incident velocity are accurately predicted. It can be seen that there is excellent agreement in terms of specimen appearance. showing (a) dishing. 3 shows the predicted deformation histories for impact speeds of 200. as a function of incident velocity. Dean et al. Dishing and discing-type failures occur at low velocities. 4 shows the experimental and predicted specimen shapes after projectile penetration. . for the impact velocities shown. Comparison between the modelled (left) and experimentally-observed (right) appearance of specimens after projectile penetration. for the times shown (after initial projectile contact). 316 and 580 m sÀ1. (ii) shear plugging. order of 600 m sÀ1. although admittedly this energy now represents just a small fraction of the kinetic energy of the incident projectile.1254 J. 3. 4. it constitutes over 50% of the total energy absorbed. / International Journal of Impact Engineering 36 (2009) 1250–1258 Fig. just above the ballistic limit (w160 m sÀ1). (b) shear plugging and (c) fragmentation modes of failure.

. Fig. with little or no dishing (but with some petalling). compared with the behaviour at low impact velocities – see below). 6. 8. Of course. It is of interest to consider the strain rates experienced by different regions while deformation takes place.J. The high speed photograph shown in Fig. the assumption that all fragments acquire a velocity equal to the residual velocity of the projectile is a rather crude one. for three increasing projectile velocities. 3(b) indicate that the energy absorbed within the specimen (by plastic deformation. Fig. Plugging failure occurs at higher impact velocities (>w250 m sÀ1). which is probably responsible for some of the scatter in the experimental data in Fig. so the agreement exhibited here constitutes evidence of the reliability of the model which is independent of the comparison shown in Fig. These experimental data were obtained from measured mass loss data (see Table 2). but in general it is expected to be a fair approximation and this is consistent with the level of agreement between predicted and experimental data in Fig. 3(b) that. 4. It’s fairly clear that plug and fragment velocities can be in a similar range to the residual projectile velocity. for a case in which plugging occurs. the strain Fig. Predicted histories of projectile and plug velocity. 8 between model predictions and experimental data for the fragment kinetic energy. obtained by measurement of the lost mass and based on the assumption that all fragments emerge with the same velocity as the projectile. This is not unexpected. 7. failure tends to occur by a combination of extensive petal formation and fragmentation.3. Dean et al. Comparison between the model predictions for the aggregate kinetic energy of ejected (single or multiple) fragments and experimental data.2. with associated bulging. Deformation mechanisms and local strain rates The predictions shown in Fig. 4. 5. after penetrating the specimen with an incident velocity of 592 m sÀ1. fracture etc) shows a tendency to reach a plateau as the projectile velocity is increased. 8. for an impact velocity of 316 m sÀ1. and indeed represents over half of the total for very high incident velocities (w600 m sÀ1). the kinetic energy associated with these fragments can constitute a significant proportion of the total energy absorbed by the specimen. for an impact velocity of 316 m sÀ1. visible in flight. showing the ejection of a plug following plate perforation at an incident velocity of 316 m sÀ1. since the specimen volume within which plastic deformation occurs does not change much with increasing projectile velocity (indeed. 9 shows. A sequence of high speed video images. 6 confirms that fragments can be seen emerging after penetration of the specimen and the mass loss data shown in Table 2 confirms that this is occurring. 7 shows predicted velocity histories for the projectile and for the plug which is ejected by it. Kinetic energy of ejected plugs or fragments It is clear from the predictions shown in Fig. for incident velocities sufficient to create plugging or fragmentation failure. 8. High speed photograph taken during projectile emergence. as well as the projectile. and there is a limit to the levels of plastic strain which can be generated within this volume. A sequence of high speed video images (Fig. Fig. At higher impact velocities (>w400 m sÀ1). 3(b). deflection and membrane stretching (tensile tearing) of the sheet. 5) show plug ejection. with several ejected fragments. it tends to fall. / International Journal of Impact Engineering 36 (2009) 1250–1258 1255 Fig. the proportion of the plastic work absorbed within the specimen. as a function of the incident projectile velocity. A comparison is presented in Fig. Fig. as a function of the strain rate at which this plastic deformation occurred. Naturally. characterised by the removal of a plug of material with a diameter similar to that of the projectile (often slightly smaller for thin plates).

9 that these range up to about 3 Â 104 sÀ1. Predicted work of deformation distributions over the range of strain rate. while straining in neighbouring material will also tend to be curtailed at a relatively early stage as the plug becomes detached. no doubt that it is preferable to use experimental stress–strain data obtained over a wide range of imposed strain rate. Naturally. Nevertheless. the constitutive relation for that strain rate (see Fig. 1) that. It can be seen that this has a peak value at the ballistic limit. rather than simply trying to identify the most reliable constitutive relations without reference to the effects responsible for them. it must be borne in mind that these predictions would themselves be affected if a different strain rate sensitivity were to be incorporated into the constitutive equations employed. there is actually a dip over the range of 300–400 m sÀ1. In fact. for a velocity of 600 m sÀ1. with a range of (sub-sonic) incident velocities (up to 600 m sÀ1). falls off as the velocity increases within the regime of dishing and discing (and the dishing becomes more localised). The tests were simulated using the Explicit FEM code in ABAQUS/CAE. These changes in mechanism are also reflected in the curve shown in Fig. remains low while plugging failure occurs and rises slightly as fragmentation replaces plugging. Absorbed energies were obtained experimentally from incident and emergent projectile velocities. according to the Johnson–Cook formulation for this material. There’s Fig. It’s also of interest to note the absolute values of typical strain rates created during this kind of deformation. Dean et al. This is the approximate range over which plugging occurs – see Fig. the FEM predictions presented here do appear to capture most of the main features of the response of this particular material to this type of ballistic loading. In order to investigate further and more deeply. / International Journal of Impact Engineering 36 (2009) 1250–1258 Fig. some interesting effects can be identified on examining these data more closely. Predicted dependence on incident velocity of (a) characteristic (mean) strain rate within the specimen throughout the process and (b) volume of material within which plastic deformation occurred. due to the lack of literature data relating to strain rates above 104 sÀ1. substantial strain rate hardening occurs under these experimental conditions. (a) Thin (0. which is clearly not really appropriate or satisfactory. It can be seen in Fig. It’s certainly plausible that the onset of this failure mechanism should lead to a reduction in the average strain rate. although obtaining these data can be time-consuming and difficult.4 mm) plates of 304 stainless steel were perforated by spherical steel projectiles. for a projectile velocity of 200 m sÀ1.2. The Johnson and Cook plasticity algorithm was . 5. since deformation rates will tend to be relatively low within a plug being pushed by a projectile. Of course. showing how the proportion of the total work done within a specimen (subjected to impact by a projectile with a given velocity) occurs within a series of bins covering the strain rate range. It’s of interest to note (Fig. and up to about 3 Â 105 sÀ1.1256 J. while there is a trend for this strain rate to increase as the incident velocity goes up. it would be advisable to bring the micro-mechanisms of deformation clearly into the picture. strain rates of w105 sÀ1 generate stress level enhancements (compared to the quasi-static case) by factors of at least about 2 – i. Conclusions The following conclusions can be drawn from this work. For example.e. 8. which shows how the volume of material subjected to plastic deformation changes as the projectile velocity increases. Fig. rates at which most of the deformation takes place tend to rise as the projectile velocity is increased.3). 9. 10. It can be seen that. 10(b). However. 1) was employed in the present work for all higher strain rates as well. 10(a) shows the predicted dependence on projectile velocity of the characteristic (mean) strain rate within a specimen (see Section 3. but these calculations certainly give an indication of the nature of the deformation taking place within the specimen under this type of loading. these strain rate values will tend to vary with projectile size and sheet thickness (and also with sheet material).

International Journal of Plasticity 2006. [24] Yungwirth CJ. Hopperstad OS. Hopperstad OS. The model also captured the transitions in failure mode. to plugging failure at around 300–400 m sÀ1 and a fragmentation process at higher velocities. critical plastic strain fracture criterion. (c) A study has been made of the strain rates at which the deformation takes place within specimens of this type. [16] Borvik T. The DSTL contract number under which part of the work was done is RD033-3169. Johnson W.32:35–64. Necking and radial cracking around perforations in thin sheets at normal incidence. Deshpande VS. Part 1: experimental study. Malo KA. Journal of Applied Physics 1955. Ballistic penetration of steel plates. Wadley HNG. Hashemi J. [12] Teng X. Dynamic failure of metallic pyramidal truss core materials – experiments and modelling. Acknowledgements Financial support has come from DSTL (for JD). 378. Appendix. hemispherical and conical noses. / International Journal of Impact Engineering 36 (2009) 1250–1258 1257 employed. [27] Lee SC. temperatures and pressures. Ruiz C. Wierzbicki T. and from EPSRC (for CSD). 2006. Numerical simulation of plugging failure in ballistic penetration. Adiabatic shear localisation in the dynamic punch test.27:19–35. Mason JJ. For example. in general. A B C cp D1–D5 E m n R T t U V v a 3 3 ðd3=dtÞ (d3/dt)0 r s sY sUTS s Johnson–Cook material parameter (Pa) Johnson–Cook material parameter (Pa) Johnson–Cook material parameter (–) Specific heat (J kgÀ1 KÀ1) Johnson–Cook fracture constants (–) Young’s modulus (Pa) Johnson–Cook material parameter (–) Johnson–Cook material parameter (–) Yield strength ratio (–) Temperature (K) Time (s) Energy (J) Velocity (m sÀ1) Volume (m3) Proportion of deformation work converted to heat (–) Strain (–) Equivalent strain (–) Equivalent strain rate (sÀ1) Reference strain rate (sÀ1) Density (kg mÀ3) Stress (Pa) Yield stress (Pa) Ultimate tensile strength (Pa) Effective stress (Pa) . [13] Zukas JA. The strain rates that are operative while most of the plastic deformation occurs during ballistic impact fall mostly in the range of 102–104 sÀ1 for an incident velocity of 200 m sÀ1 and 103–105 sÀ1 for an incident velocity of 600 m sÀ1.J. 1st ed. International Journal of Impact Engineering 1999. [23] Chou PC. International Journal of Impact Engineering 1996. Petalling of thin metallic plates during penetration by cylindro-conical projectiles.and hemispherical-nosed projectiles. Langseth M. O’Connor JH.61: 1816–25. Giles Aldrich-Smith and Neil Bourne (AWE). Hutchinson JW. Engineering Fracture Mechanics 2006. [3] Corbett GG. Dislocation-mechanics-based constitutive relations for material dynamics calculations. European Journal of Mechanics A – Solids 2008. Metallurgical and Materials Transactions A – Physical Metallurgy and Materials Science 2004. The authors are grateful for useful discussions with a number of researchers with an interest in this area. It is shown that. ABAQUS/explicit user’s manuals. Hopperstad OS. Petalling of plates under explosive and impact loading.22:935–54. Reid SR. [8] Lee YW. There is currently a shortage of such data for most materials. Impact response of sandwich panels with a pyramidal lattice core. Cadt CM.26(10):1248–53. [15] Zerilli FJ. Engineering Fracture Mechanics 1985. Khan MA. [20] Wright TW. Chou A. International Journal of Solids and Structures 1983.22:2118–45.11:305–21.22:855–86. [6] Atkins AG. [5] Landkof B. Hixson RS. International Journal of Impact Engineering 1991. International Journal of Impact Engineering 1999. Gray III GT. [19] Borvik T. Armstrong RW. [4] Wierzbicki T. Goldsmith W. Wierzbicki T. [26] Krafft JM. International Journal of Impact Engineering 2006. Nomenclature S q Stress triaxiality (–) Dimensionless temperature (–) References [1] Corran RSJ. Ballistic impact: the status of analytical and numerical modelling. International Journal of Impact Engineering 2008. Espinosa HD. it has been shown that the failure mode is important. At strain rates of this magnitude. version 6.1(1):3–22. the contribution to the total energy absorbed represented by the kinetic energy of escaping small fragments of the specimen becomes highly significant (>50% of the total).4(1):43–58. Experimental and numerical study on the perforation of AA6005-T6 panels. Zakraysek AJ. Shadbolt PJ. [17] Borvik T. 2002. during ballistic impact. International Journal of Impact Engineering 2002. there is a tendency for strain rates to increase at higher impact velocities. Berstad T.18(2): 141–230. International Journal of Impact Engineering 1999. Hopperstad OS. Rutherford RW. [18] Teng X. Impact loading of plates and shells by free flying projectiles. Wierzbicki T. Hopperstad OS. [28] Maloy SA. [22] Roessig KM. Liu JH.22(2–3):95–395. p. the sequence of events during plugging failure is such that the mean strain rate tends to fall somewhat as the velocity becomes high enough for it to become dominant. Journal of Applied Physics 1987. International Journal of Impact Engineering 2002. from dishing at low velocities.21(1):31–48. [2] Anderson CE. Langseth M. towards the upper end of the velocity range investigated. International Journal of Impact Engineering 2005.32:1921–44. Perforation of 12 mm thick steel plates by 20 mm diameter projectiles with flat. Experimentation and finite element simulation of adiabatic shear bands in controlled penetration impact. International Journal of Solids and Structures 1999. strain rates. Dodd B. Eriksson M. Computing Systems in Engineering 1993.35:920–36.38:6241–64. Cambridge University Press. Transition of failure modes in round-nosed mass-tobeam impact. Langseth M. Bodner SR. Sekhon GS. Iqbal MA. [7] Borvik T. Perforation of 12 mm thick steel plates by 20 mm diameter projectiles with flat. Evaluation of six fracture models in high velocity perforation. Berstad T. Some common problems in the numerical modeling of impact phenomena. Rogers HC.73:1653–78. Adiabatic shear localization. [10] Gupta NK. The physics and mathematics of adiabatic shear bands. International Journal of Impact Engineering 1998. [21] Bai Y.27:37–64.15(2):263–83. Barthelat F.6. Impact loading of plates – an experimental investigation. [25] ABAQUS. This was coupled with a strain rate-dependent. International Journal of Impact Engineering 2005. including Richard Jones (DSTL) and Andrew Wallwork.35(9):2617–24. Berstad T.7(1):9–35. International Journal of Impact Engineering 1983. hemispherical and conical noses part 2: numerical simulations. Non-ideal projectile impact on targets. Clausen HA.21(3):245–66.31:1253–76. part 2: numerical simulations. [14] Johnson GR. [9] Goldsmith W. Part 1: dishing. Experimental and numerical studies on the behaviour of thin aluminium plates subjected to impact by blunt. International Journal of Impact Engineering 1988. Cook WH. Langseth M. The influence of explosive-driven ‘‘Taylor Wave’’ shock pre-straining on the structure/ property behaviour of 304 stainless steel. Surface friction in ballistic penetration. While. Oxford: Pergamon Press. this steel exhibits substantial strain rate hardening and it’s clearly important to have reliable input data concerning this effect if these impact processes are to be realistically modelled. Langseth M. Fracture characteristics of three metals subjected to various strains. International Journal of Plasticity 1999. 1992. [11] Borvik T. Malo AK. Fracture prediction of thin plates under localised impulsive loading. (b) The predicted amount of energy absorbed by the specimen was found to agree well with the experimental data over the complete range of projectile velocity.21(7):521–39. Dean et al.

Deshpande VS. Part 2. . PhD thesis.38:176–91. of Engineering. University of Cambridge. Numerical study. Berstad T. Effect of strain rate on stress–strain behaviour of alloy 309 and 304L austenitic stainless steel. [31] Dey S. On the influence of fracture criterion in projectile impact of steel plates. Borvik T. On the influence of stress triaxiality and strain rate on the behaviour of a structural steel. Impact response of lightweight sandwich panels. Metallurgical and Materials Transactions A 2006.22:15–32. [34] Radford DD. Fleck NA. Dean et al. London: Oxford University Press. / International Journal of Impact Engineering 36 (2009) 1250–1258 [32] Zhou D.1258 J. [33] Lichtenfeld JA. van-Tyne CJ. Computational Materials Science 2006.37A:147. European Journal of Mechanics A – Solids 2003. McShane GJ. Mataya MC. Cambridge: Dept. 1950. International Journal of Solids and Structures 2006. Hopperstad OS.43:2243–59. 2005. The response of clamped sandwich plates with metallic foam cores to simulated blast loading. The mathematical theory of plasticity. Hopperstad OS. [29] Borvik T. Langseth M. [30] Hill R.